Fix 6604

apache · Jul 26, 2024 · 190d377 · 190d377
1 parent b93b098
commit 190d377
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 2 deletions.
diff --git a/...scala/org/apache/spark/sql/execution/datasources/utils/MergeTreePartsPartitionsUtil.scala b/...scala/org/apache/spark/sql/execution/datasources/utils/MergeTreePartsPartitionsUtil.scala
@@ -166,7 +166,7 @@ object MergeTreePartsPartitionsUtil extends Logging {
         partition =>
           partition.files.map(
             fs => {
-              val path = fs.getPath.toString
+              val path = fs.getPath.toUri.toString
 
               val ret = ClickhouseSnapshot.pathToAddMTPCache.getIfPresent(path)
               if (ret == null) {

diff --git a/.../apache/spark/sql/execution/datasources/v1/clickhouse/MergeTreeFileFormatDataWriter.scala b/.../apache/spark/sql/execution/datasources/v1/clickhouse/MergeTreeFileFormatDataWriter.scala
@@ -306,7 +306,8 @@ abstract class MergeTreeBaseDynamicPartitionDataWriter(
       releaseCurrentWriter()
     }
 
-    val partDir = partitionValues.map(getPartitionPath(_))
+    val partDir =
+      partitionValues.map(getPartitionPath(_)).map(str => new Path(str).toUri.toASCIIString)
     partDir.foreach(updatedPartitions.add)
 
     val bucketIdStr = bucketId.map(id => f"$id%05d").getOrElse("")

diff --git a/...ouse/src/test/scala/org/apache/gluten/execution/GlutenClickHouseMergeTreeWriteSuite.scala b/...ouse/src/test/scala/org/apache/gluten/execution/GlutenClickHouseMergeTreeWriteSuite.scala
@@ -2022,5 +2022,33 @@ class GlutenClickHouseMergeTreeWriteSuite
          |""".stripMargin
     runTPCHQueryBySQL(6, sqlStr) { _ => }
   }
+
+  test("test mergetree with partition with whitespace") {
+    spark.sql(s"""
+                 |DROP TABLE IF EXISTS lineitem_mergetree_partition_with_whitespace;
+                 |""".stripMargin)
+
+    spark.sql(s"""
+                 |CREATE TABLE IF NOT EXISTS lineitem_mergetree_partition_with_whitespace
+                 |(
+                 | l_returnflag    string,
+                 | l_linestatus    string
+                 |)
+                 |USING clickhouse
+                 |PARTITIONED BY (l_returnflag)
+                 |LOCATION '$basePath/lineitem_mergetree_partition_with_whitespace'
+                 |""".stripMargin)
+
+    spark.sql(s"""
+                 | insert into table lineitem_mergetree_partition_with_whitespace
+                 | (l_returnflag, l_linestatus) values ('a A', 'abc')
+                 |""".stripMargin)
+
+    val sqlStr =
+      s"""
+         |SELECT * from lineitem_mergetree_partition_with_whitespace
+         |""".stripMargin
+    runSql(sqlStr) { _ => }
+  }
 }
 // scalastyle:off line.size.limit